The Ledger · Issue One, October 2026
What the Record Will Support
22 of 77 compound names return a US drug application record. 55 return none. For anyone underwriting an asset here, the question is less what a compound does than what its paperwork will carry.
Read with Feature I on registered development and Feature II on where the regulatory record actually points.
Two piles of paper
A query of the US drug application register on 2026-10-09, run once per compound name across this corpus, returns application records for 22 of 77 names. 55 return none.
Among the 22, the counts are themselves uneven: Leuprolide 21, Desmopressin 20, Octreotide 17, Liraglutide 11, Vasopressin 10, Somatropin 5. A high count is a history of applications under a name over decades, often across several sponsors and several forms. It is not a quality score and it is not a claim about any product.
One pile of names has a regulatory file. The other does not.
The division is worth holding onto because it is the cleanest partition in this issue. One pile of names has a regulatory file. The other does not. Everything commercial downstream, what a lender will lend against, what an insurer will write, what a buyer will accept in a data room, reads very differently on the two piles, and no amount of scientific interest moves a name from one to the other.
| Compound name as queried | Application records |
|---|---|
| Leuprolide | 21 |
| Desmopressin | 20 |
| Octreotide | 17 |
| Liraglutide | 11 |
| Vasopressin | 10 |
| Somatropin | 5 |
| Triptorelin | 4 |
| Oxytocin | 3 |
| Teriparatide | 3 |
| Exenatide | 2 |
| Tirzepatide | 2 |
| Afamelanotide | 1 |
| Bremelanotide | 1 |
| Dulaglutide | 1 |
| Elamipretide | 1 |
| Glutathione | 1 |
| Lixisenatide | 1 |
| Orforglipron | 1 |
| Pramlintide | 1 |
| Semaglutide | 1 |
| Setmelanotide | 1 |
| Tesamorelin | 1 |
What a documentary record of this shape holds
The corpus this publication draws on carries 77 compound records across 11 classes, 738 cited source records and 902 indexed study rows. Each compound also has a profile that describes which kinds of record the corpus holds for it.
That profile carries a derived tier, and the tier needs its caveat printed in the same breath as the number, every time. A tier here is counted from study models and identity fields. It is labelled as inferred in the data, it carries its own basis string, and it is a description of what the corpus holds about a compound. It is not an assessment of clinical efficacy, of safety, of study quality or of scientific consensus, and it must never be read as one.
With that said, the distribution is the business fact. Of 77 profiles, 23 sit in the tier for compounds where the corpus has indexed human trial records, and 24 in the tier where it has indexed only animal and laboratory records. A further 13 are preparations, proteins or mixtures for which no defined small-molecule identity record was returned at all, which means the most basic identity fields are unsourced rather than wrong.
An asset built on a name in that last group is an asset whose data room begins with an absence.
| Tier, as the data labels it | Compounds |
|---|---|
| Animal and in vitro records only | 24 |
| Human trial records indexed | 23 |
| Limited human records | 16 |
| Preparation, protein or mixture | 13 |
| Study model not established | 1 |
The cost of being careful
There is a line item in this sector that rarely appears in anyone’s model, and this publication can price it from its own operation.
Of the 77 evidence profiles in the corpus, 50 are classified as high impact, meaning the subject is one where a reader could mistake curated research for medical guidance. Every one of those 50 is recorded as awaiting qualified human review, and until that review is recorded the corresponding reference pages are built but withheld from indexing.
50 of 77 entries, in other words, are content that exists and cannot be published. Not because any of it is wrong, but because publishing it responsibly requires a qualified reviewer, and a qualified reviewer is a recurring cost with a queue attached.
Any operator publishing in this category faces the same arithmetic, whether or not they have written it down. The compounds with the most attention are, almost by definition, the ones that attract the strictest review. The economics of a content position here are therefore inverted from most sectors: the pages that would earn the most are the pages that cost the most to clear, and the cheap pages are cheap because nobody is looking for them.
The desk’s reading
The argument, again stated as the desk’s own.
In most sectors a technical asset is underwritten on what it does. Here it is underwritten on what can be shown. The two piles in the first chapter are not a ranking of compounds by merit; they are a ranking of compounds by the amount of third-party documentation an outside party can pull without taking anyone’s word for anything. That is the variable that moves a valuation conversation, and it is almost entirely independent of pharmacology.
The practical test is a short one. For a given name, can a reader assemble, from public registers alone, an application history, a registered development record, and a cited literature base? For the 22 names with an application record the answer is usually yes. For much of the remainder the answer is no, and the honest description of such an asset is a commercial position supported by demand and a supply route, which is a real kind of business but a different one.
Feature IV takes the demand half of that sentence and asks whether it is carrying as much weight as people think.
Sources, and the queries that produced every figure above
- Drugs@FDA application register, queried through the openFDA drugsfda endpoint. U.S. Food and Drug Administration. Retrieved 2026-10-09. Read from this project’s own compound records, which store the result of a single automated query per compound name on the stated date together with the wording the query returned.https://api.fda.gov/drug/drugsfda.json?search=openfda.generic_name:"<name>"&limit=5
- Peptide research corpus: 77 compound records and 77 evidence profiles. This project’s validated dataset, data/compounds and data/evidence. Counts are taken directly from the dataset files. A derived tier on an evidence profile is labelled INFERRED in the data and carries its own basis string, which is reproduced wherever a tier is printed.
- ClinicalTrials.gov Data API, version 2, studies endpoint. U.S. National Library of Medicine, National Institutes of Health. Retrieved 2026-10-13. One query per corpus compound name. totalCount is the number of study records the registry returns for the term on the retrieval date. The full result set, one row per compound with the URL that produced it, is published in data/quarterly/clinicaltrials.json.https://clinicaltrials.gov/api/v2/studies?query.intr=<compound name>&countTotal=true&pageSize=1